Watching TikTok and Instagram deactivates the cognitive control network: Study

Millions of people finish short video after short video every day; a new brain-scan study shows that the very act of finishing a clip they like temporarily quiets the brain regions that normally help them stay focused and weigh longer-term goals. When people watch a short video they enjoy enough to finish, two brain regions involved in cognitive control show significant deactivation. That is the central finding of a new study from Zhejiang University, published in NeuroImagein January 2026. Using functional MRI alongside proton magnetic resonance spectroscopy (¹H-MRS), the research team examined 56 young adults while they freely watched short video clips inside an MRI scanner. Both the dorsal anterior cingulate cortex (dACC) and the dorsolateral prefrontal cortex (dlPFC) showed reduced activity specifically when participants watched clips they liked enough to view to completion.

Cognitive control helps people balance immediate pleasures against longer-term goals, and impairments in this system are linked to conditions such as depression, anxiety, ADHD, and addiction. Short-video platforms present rapid, algorithmically curated streams that are built for continuous, low-effort consumption. Prior behavioral research has tied both internet addiction and smartphone addiction to weaker self-control, and separate neuroimaging work has documented disruptions to reward and cognitive-control circuits in people with behavioral addictions. Despite this, few studies had directly tested whether the act of watching entertaining short videos itself suppresses the brain’s cognitive control regions. The Zhejiang University team set out to answer that question, along with a second one: what neurochemical factors might explain why this suppression varies from person to person?

The dACC and dlPFC form the core of the brain’s cognitive control network. The dACC contributes to conflict monitoring, reward-based decisions, and effort evaluation, and it typically activates during demanding tasks such as the Stroop or Go/No-Go test. The dlPFC, which connects structurally to the dACC, carries out top-down control once the dACC has flagged a need for it. Earlier work from the same lab had already shown that passively viewing personalized short videos suppresses the dACC, regardless of whether the content was algorithmically recommended or generic. Separately, prior neurochemical research has linked resting-state glutamate concentrations in the anterior cingulate cortex to stronger task-related brain activation, while GABA concentrations have been associated with reduced activation in some contexts. However, these metabolite-to-activity relationships have proven inconsistent across different types of cognitive tasks.

The team recruited 66 volunteers and excluded 10 for excessive head motion or low-quality spectroscopy data, leaving a final sample of 56 participants (37 men and 19 women, average age 23.3). All participants reported some existing experience with short-video apps. Before scanning, the researchers measured resting-state glutamate and GABA concentrations in each participant’s dACC using a MEGA-PRESS spectroscopy sequence, with a matched voxel in the visual cortex serving as a control region.

During the scan, participants watched two six-minute blocks of short clips drawn from a library of 160 videos spanning five categories: single-person actions, multi-person interactions, pets, game scenes, and natural scenery. Average clip length was 23.7 seconds. Participants could press a button at any time to skip to the next clip. Based on viewing behavior, each video was classified as “liked” (watched to the end), “disliked” (skipped before the halfway point), or a third category for clips skipped after the halfway point. On average, participants watched about 19 liked videos and 36 disliked videos per session. The study had been preregistered in September 2024, and the analysis plan set a Bonferroni-corrected significance threshold to account for multiple comparisons.

Both the dACC and dlPFC deactivated significantly below baseline while participants watched liked videos. During disliked videos, the pattern diverged: dACC activity stayed close to baseline, while dlPFC activity remained suppressed, though less severely than during liked viewing. Directly comparing conditions confirmed that both regions showed significantly greater suppression during liked video viewing than during disliked viewing. The visual cortex, used as a control region, activated during both video types with no difference between them, indicating the deactivation pattern was specific to cognitive control regions rather than a general effect of watching video.

Resting-state dACC glutamate concentration also mattered. Independent of GABA, higher dACC glutamate predicted less suppression of dACC activity during both liked and disliked viewing, and less suppression of dlPFC activity during disliked viewing. The link between glutamate and dlPFC activity during liked viewing did not reach statistical significance. GABA concentration, meanwhile, correlated significantly with activation in the visual-cortex control region but showed no significant relationship with dACC or dlPFC activity.

Functional connectivity between the dACC and dlPFC increased above baseline during both liked and disliked viewing, and this increase was significantly stronger during liked videos. Connectivity between the dACC and the visual-cortex control region showed no significant deviation from baseline in either condition. Neither glutamate nor GABA concentration significantly predicted the strength of dACC-dlPFC or dACC-V1 connectivity.

The researchers caution against reading the deactivation as evidence of impaired cognitive function. As they put it in the paper, “this deactivation should not be interpreted as a failure of cognitive control capacity.” Instead, the team argues the pattern likely reflects an adaptive shift toward low-effort, automatic processing during passive, low-conflict viewing. They note that participants remained actively engaged throughout the task, skipping disliked clips on roughly 57 percent of trials, which the authors say points to sustained evaluation rather than disengagement or mind-wandering.

The authors connect their results to prior work on “flow” states associated with other immersive media, such as films and video games, which similarly show reduced prefrontal monitoring during periods of absorbed attention. For the connectivity findings, the researchers propose that stronger dACC-dlPFC coupling during liked videos may reflect shared modulation of both regions, potentially via input from the amygdala, rather than the heightened conflict-driven engagement that typically accompanies increased connectivity in cognitive control tasks. They frame this as a departure from the classical view that tighter dACC-dlPFC coupling always signals more active cognitive control.

The authors list several limitations directly in the paper. They did not measure neurotransmitter concentrations in the dlPFC itself, only in the dACC, which limits conclusions about that region’s neurochemistry. The study did not test whether amygdala activity causally drives the deactivation it observed, despite proposing that link as an explanation. Participants’ habitual or problematic short-video use was not formally assessed with a validated addiction scale, so the findings cannot be tied to compulsive use patterns. Videos were classified as liked or disliked based only on whether participants watched them to completion, which the authors acknowledge does not fully separate genuine preference from curiosity or other factors. The study also does not address whether repeated deactivation of these regions carries any longer-term effect on cognitive function, since it captured only a single session. Finally, the sample consisted of young, healthy adults with more men than women, and the authors note that documented sex differences in glutamate and GABA metabolism mean the results may not generalize to other age groups or populations. Because the design was correlational, the paper does not establish whether glutamate levels cause the observed differences in brain activity or simply track alongside them.

Reference:

Hong, T., Su, C., Zhou, H., Geng, F., & Hu, Y. (2026). Brain activity inhibition during short video viewing: Neurochemical insights. NeuroImage, 327, 121722. https://doi.org/10.1016/j.neuroimage.2026.121722